RDP 2010-01: Reconciling Microeconomic and Macroeconomic Estimates of Price Stickiness 5. Macroeconomic Implications

We have shown that ignoring heterogeneity in the frequency of price resetting when estimating the NKPC leads to the aggregate Calvo parameter being considerably overstated, and that this bias is greater if roundabout production is a characteristic of the economy. To examine the macroeconomic consequences more broadly, we can compare the impulse response functions (to the same sized shocks) from the full model to those of a modified baseline model (Figure 4). The modification is that the forward-looking Phillips curve is replaced with parameter values similar to those estimated using GMM, namely β = 0.99, θ = 0.77, and ω = 0.17 (Table 6). This model is akin to the ubiquitous three-equation New-Keynesian model.

Figure 4: Impulse Responses from GMM Estimates

Although the two models are quite different, the impulse response functions are surprisingly similar. The modified model appears to be able to mimic the effects of heterogeneity and roundabout production in the full model with the combination of a larger Calvo probability and indexation. The main difference between the Phillips curves of the two models lies in their interpretation. Taken literally, one would falsely interpret the GMM estimates as implying that prices change every seven quarters on average and that roughly 20 per cent of firms choose to index their prices to past inflation. Instead, the results should be interpreted just as a set of parameters which can replicate a more realistic and complex data-generating process.[16]


Note that these parameters were obtained through conventional GMM estimation of the NKPC, rather than using Carvalho's (2006) approach of matching the impulse response functions of a model excluding heterogeneity to one including it. [16]